Newsletter

Please share a bit about your professional background and experience.

I graduated from Istanbul Technical University, where I studied AI and Data Engineering for my bachelor’s degree. During my bachelor’s studies, I completed an internship as a Machine Learning Engineer in Berlin, Germany, and worked as a Data Scientist at Turkish Airlines Technology during my senior year. Right now, I am pursuing my PhD in Computer Science.

What role do you play at COSMOS?

I am a graduate research assistant at the COSMOS Research Center and am helping Prof. Agarwal, who is leading the projects on collective action and toxicity.

What attracted you to join Prof. Agarwal’s research and COSMOS Center? What aspects of Prof. Agarwal’s vision, mission, and the culture he has fostered at COSMOS stood out to you and why?

As an individual who wanted to pursue further research after completing my bachelor’s degree and who loves storytelling, I felt that social computing was a good match for me. Prof Agarwal is a world-renowned expert in Social Computing; therefore, it was natural for me to join the COSMOS Research Center that he is leading and learn how to conduct research from him. At COSMOS, we examine how social media drives human behavior from several perspectives. The significant computational and interdisciplinary research infrastructure that Prof. Agarwal has established through his numerous highly prestigious grants enables our research to run much more smoothly.

How do you anticipate your role at COSMOS helping your growth on both a personal and professional level? Are there any specific skills or experiences you’re looking to gain?

As a graduate research assistant under Prof. Agarwal’s tutelage, I am looking forward to learning every aspect of conducting research, including literature surveys, helping conduct experiments, and helping write papers. My goals are to grow as a researcher, improve my soft and hard skills, such as learn foundational theories in computational social science, apply them to novel research ideas, and improve my communication of these findings.

From your experience, what tips, insights, or advice would you share with someone starting a new role at COSMOS?

Prof Agarwal has fostered a professional, collaborative, real-world problem-solving environment and assembled a great team and resources at the COSMOS Research Center. He has an infectious energy and a hard-working work ethic. He is highly regarded in the global research community. We all should learn from him and contribute to the scientific community. We have a great team at COSMOS, so engage and learn from others – and enjoy while doing so!

If you could share a meal with any historical figure or fictional character, who would it be, and what would you want to talk about and want to learn from them?

I would say John McCarthy, also known as the “father of AI”. The reason is that I really enjoyed my bachelor’s education and believe that the foundational ideas that underpin today’s development are crucial to understand. I would like to discuss what McCarthy would think if he saw the current state of AI, especially after late 2022, when Large Language Models began attracting massive audiences.

Insert test here…

COSMOS at UA Little Rock continues to push the boundaries of socio-computational research, unveiling three groundbreaking studies at the prestigious and highly interdisciplinary 18th International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction (SBP-BRiMS). Hosted at Carnegie Mellon University, this elite forum brings together global leaders in computer science and social behavioral modeling to address the world’s most pressing digital threats.

The study, entitled “Studying Emotional and Trust-building Effects of Symbolic Communication on YouTube,” analyzes Taiwan’s information campaigns. LLMs were used in the research to reveal that cultural symbols drive significantly higher emotional engagement and trust than non-symbolic content, proving that visual semiotics are critical to digital political discourse.

Another study, entitled “Structure, Semantics, and Attraction: Analyzing Homophily in Recommender Networks,” used Focal Structure Analysis (FSA) to identify dense network subgraphs on YouTube that act as behavioral “traps.” These structures exhibit high topical uniformity, effectively narrowing user exposure and reinforcing ideological silos.

The third study, entitled “Uncovering Structural Consistency in YouTube Channels,” proposed a new unsupervised framework that characterizes YouTube channels by the semantic alignment of their metadata (titles, descriptions) with actual transcripts. The study identified three distinct behavioral profiles, offering a scalable method to detect narrative misrepresentation without manual labeling.

While unique in their focus, from visual semiotics to network topology and metadata alignment, these studies collectively advance AI and Social Computing. While each paper addresses a unique facet of AI, i.e., from multi-modal signal processing to geopolitical modeling, they collectively advance our ability to secure the socio-cognitive domain. COSMOS’s contributions at SBP-BRiMS highlight a unified mission: bridging the gap between social science theory and advanced machine learning to build community resilience.

By integrating Large Language Models (LLMs) with traditional network analysis, COSMOS is building the tools necessary to understand and mitigate harmful behaviors, ensuring transparency and resilience in our increasingly complex social-cyber world. By pioneering these socio-computational approaches, COSMOS is not only advancing the state-of-the-art in AI and YouTube forensics but is also providing policymakers and technologists with the tools needed to mitigate harmful behaviors and neutralize coordinated cognitive threats in our increasingly complex digital world.

COSMOS continues to advance global scholarship in AI and network science with a new study published in the Journal of Applied Network Science, a premier, highly selective Springer Nature journal recognized for influential research at the intersection of networks, data science, and societal impact.

In this article, COSMOS researchers investigate how symbolic content, such as social, cultural, and political imagery, shapes information diffusion within YouTube’s recommendation ecosystem. Using large-scale social network analysis and advanced AI-enabled methodologies, the study compares the propagation dynamics of symbolic versus non-symbolic videos across multiple recommendation depths.

Key findings reveal that symbolic content is structurally advantaged within recommendation networks. Videos containing symbolic cues exhibit significantly higher influence and visibility, as measured by eigenvector centrality, closeness centrality, and PageRank, indicating that such content travels faster and occupies more central positions in the network. In contrast, degree and betweenness centrality show fewer differences, suggesting that algorithmic bias emerges not merely from volume of connections but from how influence is amplified algorithmically.

The research further demonstrates that symbolic content fosters tighter communities and deeper clustering, reinforcing echo chambers and raising concerns about algorithmic amplification in adversarial information campaigns. These findings have critical implications for AI-driven platform governance and the responsible design of recommender systems.This work underscores COSMOS’ leadership in AI-powered social media analytics, combining network science, computational social science, and ethical AI to address pressing global challenges. By publishing in the Journal of Applied Network Science, COSMOS continues to shape international discourse on algorithmic transparency, bias, and resilience in online platforms. Read the full article here.

COSMOS Director Prof. Nitin Agarwal represented the University of Arkansas at Little Rock at the 17th IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM), held in Niagara Falls, Ontario, Canada, contributing to a high-profile international panel on the growing influence of large language models (LLMs) on social networking and social media.

The panel, moderated by Rokia Missaoui (Université du Québec en Outaouais, Canada), brought together leading experts from North America, Europe, and Asia, including Jean-Loup Guillaume (La Rochelle Université, France), Hongxin Hu (University at Buffalo, USA), Rasha Kashef (Toronto Metropolitan University, Canada), Kwan Hui Lim (Singapore University of Technology and Design), and Tamer Özsu (University of Waterloo, Canada).

Prof. Agarwal’s remarks focused on the transformative role of LLMs in shaping social media ecosystems, with particular attention to their impact across academia and industry. He highlighted both the positive potential—such as accelerating research, enhancing teaching and learning, and enabling advanced analytics in sectors like banking and e-commerce—and the risks, including manipulation, deception, bias, opacity, and adversarial misuse at scale.

Drawing on COSMOS’s long-standing research leadership in AI, social computing, and cognitive security, Prof. Agarwal emphasized critical challenges and design considerations for responsibly integrating LLMs into social network analysis and mining, including transparency, explainability, and human-AI collaboration. He also discussed emerging opportunities for using LLMs to query complex social media data and generate interpretable explanations, a key step toward trustworthy and actionable social media intelligence.

Participation in this panel underscores COSMOS’s growing global footprint and its role in advancing responsible, interdisciplinary AI research on the world stage.

There is a new addition to the COSMOS family! We hired a new full-time Data Engineer, Atharvaa Rane! Atharvaa shares what attracted her to COSMOS, as well as her aspirations and goals.

What is your name, and what role do you play at COSMOS?

My name is Atharvaa Rane, and I work at the COSMOS Research Center as a Data Engineer. I am working on the BlogTracker project, where I help build data pipelines, backend APIs, and AI-driven analytics for blog data. My role involves improving system architecture, enhancing data ingestion and processing workflows, and supporting machine learning–based analysis.

Can you share a bit about your professional background and experience?

I recently completed my Master’s in Computer Science from Syracuse University, where my focus was on machine learning, data engineering, and analytics. Alongside academics, I’ve worked in roles that sit at the intersection of data, engineering, and applied AI. Upon graduation, I worked at iConsult Collaborative as a Data Analyst, building Python and SQL pipelines and Power BI dashboards to automate compliance tracking and improve decision-making for real clients. I’ve also worked as a Data Engineer and AI/ML Developer Intern at ToyzElectronics, where I helped develop personalized recommendation systems and APIs for an ed-tech platform serving over 10,000 users. Across these roles, I’ve enjoyed end-to-end problem solving—from messy data ingestion to building systems that people actually use.

Which aspects of COSMOS culture or mission stood out to you and why?

What stood out most to me about COSMOS is Prof. Agarwal’s strong focus on research with real-world impact. The work led by Prof. Agarwal doesn’t stop at publishing or building models for academic publications — it’s about making it accessible to society at scale. I also value the interdisciplinary and collaborative culture fostered by Prof. Agarwal at COSMOS, where engineers, researchers, and domain experts work closely together. That balance of technical depth, curiosity, and purpose really aligns with how I want to grow as a researcher and engineer.

Are there any specific skills or experiences you’re looking to gain here?

I’m particularly excited to deepen my experience in building large-scale data pipelines and social media analysis, and applied machine learning. I want to sharpen my ability to design systems that are not just technically sound, but also interpretable, ethical, and useful for researchers and stakeholders. I’m also eager to learn more about translating research ideas into robust, production-ready tools.

Did anything during your onboarding experience surprise or stand out to you?

What stood out during onboarding was how welcoming and supportive the environment felt. I appreciated how open everyone was to questions and how clearly expectations and goals were communicated by Prof Agarwal. It made it easier to feel like I could contribute early on rather than just observe from the sidelines.

How do you anticipate your role at COSMOS helping your growth on both a personal and professional level? 

Professionally, this role pushes me to think more deeply about data beyond accuracy and performance, to consider context, impact, and responsibility. Working at COSMOS encourages me to be more thoughtful in how I design systems and interpret results. On a personal level, collaborating with researchers from different backgrounds helps me grow as a communicator and teammate, and it’s building my confidence in working on complex, open-ended problems.

From your experience, what tips, insights, or advice would you share with someone starting a new role at COSMOS?

My biggest advice would be to ask questions early and often, and not be afraid to explore ideas outside your immediate task list. At COSMOS, Prof. Agarwal has cultivated an environment that values curiosity, where initiative and active engagement in discussions enhance learning. Also, documenting your work and thinking out loud with teammates goes a long way in collaborative research environments.

If you could share a meal with any historical figure or fictional character, who would it be and what would you want to talk about and want to learn from them?

I’d choose Leonardo da Vinci. I’d want to talk to him about how he connected ideas across art, science, engineering, and human behavior so naturally, without putting them into separate boxes. I’d love to learn how he stayed curious across so many domains and how he approached learning when formal structures or tools didn’t exist the way they do today. That kind of mindset—blending creativity with analytical thinking—feels especially relevant when working on complex, interdisciplinary problems.

This edition features three of our peer-reviewed studies published in leading international conferences on social computing, focusing on information diffusion, anomaly detection under data scarcity, and intervention strategies for mitigating harmful online behavior in complex networked systems.

The first study, “Modeling Quarantine Intervention for Varied Toxic Intensities,” presented at The Fifteenth International Conference on Social Media Technologies, Communication, and Informatics (SOTICS 2025) in Lisbon, Portugal, explores how quarantine-based interventions can reduce the spread of toxic content online. By extending an epidemiological SEIQR framework, the study demonstrates that targeted, toxic intensity-aware quarantine strategies significantly outperform uniform moderation approaches while preserving user engagement. In recognition of its originality, rigor, and impact, this paper was recognized with the Best Paper Award at SOTICS 2025.

The other study, titled “Competing Narratives during Conflicts: Modeling Narrative Diffusion on Telegram in the Russia–Ukraine Conflict,” presented at the 18th International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction (SBP-BRiMS 2025) in Carnegie Mellon University, Pittsburgh, Pennsylvania, USA, investigates how competing pro-Kremlin and pro-Ukrainian narratives propagate on Telegram during an active geopolitical conflict. Using a stance-based epidemiological diffusion model, the study captures how rival narratives compete for attention and quantifies the structural and temporal factors that shape their spread, offering insights into information warfare dynamics on minimally moderated platforms.

The third study, titled “Evaluating Synthetic Data Generation Methods for Anomalous Channel Detection in Sparse Label Environments,” published in the proceedings of the 18th International Conference on Social Computing, Behavioral-Cultural Modeling, & Prediction (SBP-BRiMS 2025) in Carnegie Mellon University, Pittsburgh, Pennsylvania, USA, addresses the challenge of detecting anomalous YouTube channels when labeled data is scarce. The study systematically evaluates multiple synthetic data generation techniques and demonstrates that models trained on high-quality synthetic data can substantially improve anomaly detection performance in low-resource settings.

Prof. Agarwal stated that, “taken together, these studies highlight the effectiveness of mathematical modeling and data-driven approaches in understanding and managing complex social and information systems, particularly in high-stakes, adversarial, and data-constrained environments.”

At the 59th Hawaii International Conference on System Sciences (HICSS 2026) in Maui, Hawaii, Prof. Nitin Agarwal and the COSMOS Research Center showcased a compelling body of work that pushes the boundaries of computational social science, artificial intelligence, and digital behavior analysis. COSMOS presented four studies at the prestigious HICSS conference, two of which were nominated for the Best Paper award. These studies collectively reflect a bold vision: advancing AI systems that are not only powerful but also theoretically grounded, interpretable, and deeply attuned to the complexities of human behaviors observed in real-world settings. 

First study explores how multi-theoretical social science frameworks, when embedded in deep learning architectures, significantly enhance the prediction of mob outcomes. By grounding AI models in established behavioral theories, this work demonstrates how socially informed intelligence can outperform purely data-driven approaches when analyzing collective action and crowd dynamics. The second paper introduces a behavioral model of narrative diffusion that captures how individuals are exposed to information, develop skepticism, and ultimately switch beliefs. This framework offers critical insights into how narratives compete, evolve, and influence audiences within polarized information ecosystems.

The third study advances the detection of harmful discourse by presenting a neural-symbolic framework for toxic intent prediction, combining probabilistic soft logic with moral foundations theory. This approach enables more transparent and ethically informed AI systems capable of reasoning about intent rather than relying solely on surface-level language cues. Complementing this work, the fourth study examines YouTube channels through multi-view content similarity, revealing hidden relationships, coordination patterns, and thematic alignments across the platform, and providing a deeper understanding of influence and structure in large-scale media ecosystems.

Together, these contributions highlight COSMOS’s leadership in shaping next-generation AI systems that integrate social theory, ethical reasoning, and advanced analytics. Presented on a global and highly competitive stage at HICSS 2026, this body of work underscores COSMOS’s commitment to addressing some of the most pressing challenges of the digital age, where technology, behavior, and society increasingly intersect.

We are pleased to share our recent publication in Springer’s Journal of Social Network Analysis and Mining, titled “Modeling Polarized Information Diffusion with SEI(A)I(D)Z: A Stance-Based Epidemiological Approach.” This study addresses a key limitation in prior research on online information diffusion by moving beyond metadata-driven analyses and explicitly modeling the stance expressed in social media content. Recognizing that online discourse is often polarized, the work captures how agreement, disagreement, and skepticism interact simultaneously, offering a more comprehensive view of how competing narratives spread and evolve across digital platforms.

To support this approach, the study introduces a novel stance-based epidemiological framework, SEI(A)I(D)Z, which extends the traditional SEIZ model by differentiating between users who support and those who oppose a narrative. This refinement allows the model to capture interactions among competing viewpoints rather than treating information diffusion as a uniform process. The framework is empirically validated across three distinct contexts and platforms: COVID-19 and 5G conspiracy narratives on X (formerly, Twitter), geopolitical discourse surrounding the Russia–Ukraine war on Telegram, and Taiwanese information campaigns on TikTok during early 2024. Results from non-linear least-squares estimation consistently demonstrate improved predictive accuracy compared to baseline models.

The findings highlight the central role of stance, transmission rate, and user decision-making speed in shaping the reach and intensity of narrative diffusion. In particular, the transmission rate (β) emerges as the most influential driver of the basic reproduction number (R₀), illustrating how increased visibility and virality can accelerate narrative spread. The analysis also shows that slower transitions from exposure to stance adoption extend susceptibility, allowing narratives to persist longer. While manipulative narratives tend to propagate more rapidly, counter-narratives and debunking efforts require stronger systemic and institutional support to achieve comparable impact.

From a practical standpoint, the proposed framework offers actionable insights for policymakers, platform designers, and researchers seeking to curb harmful or misleading content. By explicitly distinguishing between agreement, disagreement, and skepticism, the model enables more targeted interventions, including algorithmic moderation, content throttling, and amplifying credible counter-narratives. Its scalability and adaptability across platforms further position it as a valuable tool for real-time monitoring and informed decision-making in digital governance.

Reflecting on the study, Prof. Nitin Agarwal noted that the “work advances understanding of polarized information ecosystems by explicitly modeling how competing narratives interact and evolve over time. He emphasized that this stance-based epidemiological approach provides a robust and interpretable framework for addressing the complex dynamics of information campaigns, counter-messaging, and skepticism, ultimately supporting more effective strategies to strengthen digital resilience and promote healthier online discourse.”

Click here to read the full article.

We are pleased to share that Prof. Nitin Agarwal represented the COSMOS Research Center at the UA Little Rock during the 46th International Conference on Information Systems (ICIS 2025), held from December 14–17, 2025, in Nashville, Tennessee. Recognized as one of the most prestigious conferences in the information systems (IS) discipline, ICIS serves as the flagship annual meeting of the Association for Information Systems (AIS), convening leading scholars, industry experts, and thought leaders from across the globe to engage with the latest advancements in information systems research.

The research, titled “Modeling Word-Level Functions in Social Movement Discourse,” explores how linguistic functions operate in online environments to mobilize participation and construct collective identity within social movements. By connecting word-level linguistic mechanisms to broader processes of digital organizing and sensemaking on social media, the study advances information systems scholarship through a novel socio-cognitive lens. The findings contribute both theoretically and methodologically, offering valuable insights and analytical tools for IS researchers, policymakers, and platform designers interested in civic engagement, digital governance, and AI-driven social computing.

Under the 2025 theme, “Achieving Digital Integration in the Age of AI,” the conference highlighted innovative research at the intersection of digital transformation and artificial intelligence. ICIS featured highly selective paper presentations, engaging panels, specialized workshops, and interactive discussions addressing the evolving challenges and opportunities shaping the future of information systems.

At the conference, Prof. Agarwal delivered a provoking presentation of the work, contributing meaningfully to ongoing conversations around computational modeling and socio-technical approaches to understanding digital ecosystems. His participation underscores COSMOS’s continued leadership in social computing, narrative analysis, and socio-cognitive security, while further strengthening UALR’s visibility and reputation within top-tier international research communities.

By publishing in ICIS, COSMOS remains part of a distinguished group of global scholars whose work informs both academic inquiry and real-world digital strategies, including AI governance, systems integration, social media dynamics, and digital trust. ICIS remains a premier venue for advancing the theory and practice of information systems with far-reaching impact.

This accomplishment reflects COSMOS’s research leadership and marks an important milestone, reaffirming the center’s role in addressing critical digital challenges of our time. We are incredibly proud of this outstanding contribution, which continues to elevate UA Little Rock’s presence on the global academic stage.